Visuospatial disorders detection in dementia using a computer-generated environment based on voting approach of machine learning algorithms

1. A visuospatial disorders detection method, comprising: presenting to a subject a three dimensional (3D) virtual reality environment in which the subject utilizes a multidirectional input device to input answers to questions and to guide an avatar on a path through the 3D virtual reality environment, wherein said multidirectional input device at least moves front, back, left and right; receiving input data for the subject generated by the subject's use of the multidirectional input device which comprises answers to questions input by the subject, coordinates and direction of the avatar relative to the path through the 3D virtual reality environment, and a time period used by the subject to guide the avatar along the path through the 3D virtual reality environment; supplying the received input data into a plurality of machine learning algorithms which utilizes correct and incorrect answers input by the subject, number of changes in direction of the avatar as the subject moves the avatar relative to the path through the 3D virtual reality environment, and the time period used by the subject to guide the avatar along the path through the 3D virtual reality environment, wherein the plurality of machine learning algorithms comprise Decision Tree Classifier, Extra Tree Classifier, AdaBoost Classifier, XGB Classifier, Gradient Boosting Classifier, Support Vector Classifier, Random Forest Classifier, Multinomial Naive Bayes, K-Neighbors Classifier, and Multilayer Perceptron; using machine learning with each of the plurality of machine learning algorithms to classify the subject into one of three classification labels selected from the group consisting of normal, demented, and mild cognitive impairment; and feeding results obtained with each of the plurality of machine learning algorithms into a system of voting to produce a final classification, wherein the system of voting comprises hard voting which predicts the final classification based on a most frequently used classification label produced by the machine learning using the plurality of machine learning algorithms.

2. The method of claim 1 wherein the input data received from the subject answering questions and guiding the avatar relative to the path through the 3D virtual reality environment represent testing of both memory and visuospatial function.

3. The method of claim 1 wherein the input data received from the subject answering questions and guiding the avatar relative to the path through the 3D virtual reality environment represent testing of each of navigation, visual memory, and memory function.

4. A visuospatial disorders detection method, comprising: presenting to a subject a three dimensional (3D) virtual reality environment in which the subject utilizes a multidirectional input device to input answers to questions and to guide an avatar on a path through the 3D virtual reality environment; receiving input data for the subject generated by the subject's use of the multidirectional input device which comprises answers to questions input by the subject, coordinates and direction of the avatar relative to the path through the 3D virtual reality environment, and a time period used by the subject to guide the avatar along the path through the 3D virtual reality environment; supplying the received input data into a plurality of machine learning algorithms which utilizes correct and incorrect answers input by the subject, number of changes in direction of the avatar as the subject moves the avatar relative to the path through the 3D virtual reality environment, and the time period used by the subject to guide the avatar along the path through the 3D virtual reality environment, wherein the plurality of machine learning algorithms comprise Decision Tree Classifier, Extra Tree Classifier, AdaBoost Classifier, XGB Classifier, Gradient Boosting Classifier, Support Vector Classifier, Random Forest Classifier, Multinomial Naive Bayes, K-Neighbors Classifier, and Multilayer Perceptron; using machine learning with each of the plurality of machine learning algorithms to classify the subject into one of three classification labels selected from the group consisting of normal, demented, and mild cognitive impairment; and feeding results obtained with each of the plurality of machine learning algorithms into a system of voting to produce a final classification, wherein the system of voting comprises soft voting which predicts the final classification based on averaging classification labels produced by the machine learning using the plurality of machine learning algorithms.

5. The method of claim 4 wherein the input data received from the subject answering questions and guiding the avatar relative to the path through the 3D virtual reality environment represent testing of both memory and visuospatial function.

6. The method of claim 4 wherein the input data received from the subject answering questions and guiding the avatar relative to the path through the 3D virtual reality environment represent testing of each of navigation, visual memory, and memory function.